Protein‐engineered molecules carrying GAD65 epitopes and targeting CD35 selectively down‐modulate disease‐associated human B lymphocytes

Summary

Type 1 diabetes mellitus is an autoimmune metabolic disorder characterized by chronic hyperglycemia, the presence of autoreactive T and B cells and autoantibodies against self‐antigens. A membrane‐bound enzyme on the pancreatic beta‐cells, glutamic acid decarboxylase 65 (GAD65), is one of the main autoantigens in type 1 diabetes. Autoantibodies against GAD65 are potentially involved in beta‐cell destruction and decline of pancreatic functions. The human complement receptor type 1 (CD35) on B and T lymphocytes has a suppressive activity on these cells. We hypothesized that it may be possible to eliminate GAD65‐specific B cells from type 1 diabetes patients by using chimeric molecules, containing an anti‐CD35 antibody, coupled to peptides resembling GAD65 B/T epitopes. These molecules are expected to selectively bind the anti‐GAD65 specific B cells by the co‐cross‐linking of the immunoglobulin receptor and CD35 and to deliver a suppressive signal. Two synthetic peptides derived from GAD65 protein (GAD65 epitopes) and anti‐CD35 monoclonal antibody were used for the construction of two chimeras. The immunomodulatory activity of the engineered antibodies was tested in vitro using peripheral blood mononuclear cells (PBMCs) from type 1 diabetes patients. A reduction in the number of anti‐GAD65 IgG antibody‐secreting plasma cells and increased percentage of apoptotic B lymphocytes was observed after treatment of these PBMCs with the engineered antibodies. The constructed chimeric molecules are able to selectively modulate the activity of GAD65‐specific B lymphocytes and the production of anti‐GAD65 IgG autoantibodies by co‐cross‐linking of the inhibitory CD35 and the B cell antigen receptor (BCR). This treatment presents a possible way to alter the autoimmune nature of these cells.

Introduction

Type 1 diabetes mellitus (T1DM) is a complex, multigenic autoimmune disease characterized by T and B cell autoreactivity leading to the destruction of pancreatic beta cells and loss of insulin secretion, followed by hyperglycemia, metabolic disturbances and generation of autoantibodies against a diverse array of self‐antigens. The complex pathogenesis of T1DM is heterogeneous and is caused by a primary defect of the immune system that begins to target molecules of pancreatic beta‐cells, insulin and insulin receptors 1, 2.

Clinical and experimental data indicate an immune‐mediated destruction of beta cells in genetically predisposed individuals, possibly triggered by exposure to certain environmental factors which are under discussion: viruses, nutrition, etc. Autoaggressive T lymphocytes, followed by specific self‐reactive antibodies, are recognized as pathological factors 3.

Current trials with experimental therapies of T1DM are mainly non‐specific (cyclosporine A, teplizumab, rituximab) and often counterbalanced by adverse effects 4, 5. Islet transplantation has been established as an effective therapeutic approach 6. However, the survival of the islet of Langerhans grafts can be disrupted by recurrent autoimmunity or by triggering of the adaptive immune response.

Autoreactive B and T cells may play diverse roles in the development of autoimmunity, but both lymphocyte types participate as amplifiers and effectors in T1DM 7, 8. Autoaggressive CD4⁺ T cells can interact with self‐peptides presenting B cells and thus stimulate B cell proliferation and differentiation to autoantibody‐producing plasmocytes 9, 10. In turn, CD8⁺ T cells can subject pancreatic beta cells to cytotoxic attack 11.

Great progress has been made recently to identify the major autoantigens in T1DM. Insulin and some membrane‐bound molecules are considered to be the main targets of autoimmune reaction 12. Glutamic acid decarboxylase 65 (GAD65) has been originally identified as a 65 kDa membrane‐bound enzyme that catalyses the conversion of glutamic acid to gamma aminobutyric acid (GABA) 13. Autoantibodies against GAD65 are considered as some of the most important for the disease development 14, 15, 16.

The potential role of B cells in the disease progression has recently attracted attention. Pathological GAD65‐specific B lymphocytes in patients with autoimmune diabetes can affect the onset of the disease. B cells not only secrete antibodies that recognize and participate in GAD65 breakdown, but are also substantial as autoantigen‐presenting cells. In T1DM, activated autoreactive B cells could break the T cell tolerance modifying T cell regulatory function 17, 18. The important role of B lymphocytes in the pathogenesis of T1DM has been highlighted by studies utilizing the monoclonal antibody rituximab, which recognizes CD20 on the surface of B cells and leads to their depletion 19.

Although the precise role of autoreactive B lymphocytes in the genesis of the disease has not yet been clarified, the targeted elimination or suppression of autoreactive B cells is a legitimate approach in the efforts to control T1DM 20. Disease progression is associated with increasing levels of GAD65‐specific antibodies. After a fine mapping of B cell epitopes on the GAD65 molecule, a number of monoclonal antibodies has been developed 21. Based on these findings, GAD65 has been indicated as one of the major antigens in autoimmune diabetes. As such, it has been tested extensively for both CD4 and CD8 T cell epitopes. More than 50 CD4 epitopes and only one CD8 epitope have been identified from T1DM patients or human leukocyte antigen (HLA)‐transgenic mice, most of them restricted to the T1DM‐susceptible alleles HLA‐DQ8 and HLA‐DR4 18, 22. Several CD4 T cell epitopes are localized within the frame of B cell epitopes, and their processing from B cells leads to blocking or inhibition of antigen presentation to the responding autoreactive T cells 21, 23. Here, we exploit the ability of some peptides comprising T cell epitopes to bind human B lymphocytes as part of engineered chimeric molecules.

A possible mechanism for B cell down‐regulation is the cross‐linking of the surface antigen‐binding receptors with the inhibitory B cell receptors by engineered protein chimeric molecules. It has already been published by Erdei et al. that the human CD35 can provide a strong inhibitory signal to human B lymphocytes 24. We have previously shown that the binding of specific epitopes to the B cell receptors and the cross‐linking with CD35 on the surface of autoreactive cells may have a strong negative effect on them 25, 26, 27, 28. The goal of our research is to selectively suppress anti‐GAD65 IgG antibody‐producing B lymphocytes from T1DM patients by chimeric protein molecules that contain a monoclonal antibody, specific to CD35 conjugated to peptide epitopes from GAD65. This treatment is expected to down‐regulate B cell autoreactivity and the production of human autoantibodies from GAD65‐specific plasmoblasts, delivering a strong inhibitory signal to them.

Materials and methods

Monoclonal antibodies

Mouse anti‐human CD35 monoclonal IgG1 antibody (clone 3D9) was prepared as described previously 29. Fluorescein isothiocyanate (FITC)‐conjugated mouse IgG1 isotype control (eBioscience, Frankfurt, Germany) was used for fluorescence‐activated cell sorting (FACS) experiments. Anti‐human CD35‐FITC (clone 3D9, kindly provided by Dr J. Prechl, Diagnosticum zrt, Budapest, Hungary), anti‐human CD19‐phycoerythrin (PE), CD19‐eFluor450, CD3‐PE/cyanin 7 (Cy7) and CD3‐PE/cyanin 5 (Cy5) antibodies (eBioscience) were used for FACS experiments. Mouse IgG1 kappa‐PE, IgG1 kappa‐eFluor 450, IgG1 kappa‐PE/Cy7 and IgG1 kappa‐PE/Cy5 isotype controls (eBioscience) were also used for FACS experiments.

Prediction of MHC binders by EpiDOCK

EpiDOCK is a server for prediction of peptides binding to HLA class II proteins 30. It uses quantitative matrices derived from docking‐based analyses of peptide–HLA class II complexes. EpiDOCK predicts binding to 23 most frequent HLA class II proteins: 12 HLA‐DR, 6 HLA‐DQ and 5 HLA‐DP. It is freely accessible via http://www.epidock.ddg-pharmfac.net. In the present study, EpiDOCK was used to predict binders to the type 1 diabetes susceptible alleles: DQ8 (DQA1*03:01/DQB1*03:02), DRB1*04:01 and DRB1*04:05. Default thresholds were used: 0·1 for DQ8 and 0.3 for DRB1*04:01 and DRB1*04:05.

Construction of chimeric molecules containing GAD65 epitopes and CD35‐binding antibody

Two selected peptides from the human GAD65 molecule – p121–140 and p270–289, as well as an irrelevant peptide, containing the same number of randomized amino acids – were synthesized using Fmoc chemistry on resin with > 96% purity from Caslo Laboratory (Lyngby, Denmark). The amino acid sequences were: p121–140 Ac‐YVVKSFDRSTKVIDFHYPNE‐Ahx‐K‐CONH₂ and p270–289 Ac‐LPRLIAFTSEHSHFSLKKGA‐Ahx‐K‐CONH₂.

The chimeric molecules were constructed by separate conjugation of peptides to anti‐CD35 (clone 3D9) antibodies using the zero‐length cross‐linking agent EDC [1‐ethyl‐3‐(3′‐dimethylaminopropyl) carbodiimide.HCl] (Sigma‐Aldrich, Taufkirchen, Germany), as described previously 28, 29, 31. Briefly, for the initiation of the protein conjugation, a lysine carrying Ahx linker was introduced to the C‐end of the peptides during the peptide synthesis. EDC was used to couple the available free carboxyl groups from the immunoglobulin backbone to the ε‐amino groups of lysine residues. The anti‐CD35 antibody and peptides were mixed at a 20‐fold molar excess of the peptides and 60‐fold molar excess of EDC. The reaction mixture was stirred overnight at 4^oC, and the excess of reagents was removed by dialysis against phosphate‐buffered saline (PBS). The resulted engineered protein molecules: a GAD65 chimera 1, a GAD65 chimera 2 (consisting of the p121–140 or p270–289 peptides, respectively) and a control chimera (consisting of the irrelevant peptide) were concentrated by ultrafiltration and used in the next experiments.

T1DM patients and healthy donors

T1DM‐diagnosed patients (n = 6) with high titers of anti‐GAD65 IgG antibodies and casual glucose concentration ≥200 mg/dl (11·1 mmol/l) were included in the study [female to male ratio 5 : 1; mean (min–max) age 20–32 years]. Eight age‐ and sex‐matched healthy volunteers served as controls and all the subjects in this study gave their informed consent. Local institutional ethics committee permission was obtained prior to sample collection.

Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized venous blood of both T1DM patients and healthy donors by Pancoll (PAN Biotech, Aidenbach, Germany) gradient separation. Isolated cells were cultured in RPMI‐1640 medium (Sigma‐Aldrich) supplemented with 10% fetal calf serum (FCS) and 4 mM L‐glutamine in 5% CO₂ at 37^oC.

Flow cytometry analysis

PBMCs from T1DM patients or healthy volunteers (1 × 10⁶ cells/ml) were washed with FACS buffer (PBS containing 2·5% FCS and 0·05% sodium azide) in a competition binding assay and incubated with the constructed chimeric molecules (at 1 µg/10⁶ cells) for 30 min at 4°C. The control samples were incubated with unconjugated anti‐CD35 monoclonal antibody (mAb) or with PBS alone. Next, the cells were washed twice and incubated with an optimal concentration of FITC‐conjugated anti‐human CD35 (clone 3D9) antibody combined with anti‐human CD19‐PE or with anti‐human CD3‐PE/Cy5 antibodies for 30 min at 4°C. Ten thousand cells from each sample were analyzed using a BD LSR II flow cytometer using the Diva version 6.1.1. software (BD Biosciences, Mountain View, CA, USA).

Enzyme‐linked immunosorbent assay (ELISA) for GAD65 epitope detection

The constructed chimeras were subjected to ELISA for epitopes recognition; 96‐well immunoplates (Maxisorp, Nunc, Roskilde, Denmark) were coated overnight at 4°C with GAD65 chimera 1, GAD65 chimera 2 or control chimera (0·5 mg/ml in PBS), and blocked with 0·1% gelatin in T‐PBS for 2 h at room temperature, followed by two washes. Prediluted (1 : 100 in T‐PBS) sera from T1DM patients or healthy donors were added onto the coated wells and the plates were incubated for 1 h at room temperature. The binding of the human anti‐GAD65 IgG antibodies to the immobilized chimeric molecules was assessed by subsequent incubation for 1 h with alkaline phosphatase‐conjugated anti‐human IgG antibody (Sigma‐Aldrich). Further, after thorough washing, an enzyme substrate (pNPP; Sigma) solution was added and the color development was measured spectrophotometrically at 405 nm.

Apoptotic assay

For detection of apoptosis, PBMCs from T1DM patients and healthy donors were isolated and cultured for 3 days (2 × 10⁶ cells/ml in complete RPMI‐1640 medium) with rising concentrations of the chimeric molecules (40, 100, 250 and 625 ng/ml) at 37°C/5% CO₂. Then the cells were collected, washed and stained with eFluor450‐conjugated anti‐human CD19 or anti‐human CD3‐PE/Cy7 antibodies. The levels of apoptosis of gated B or T cell populations was evaluated within 15 min by flow cytometry (BD LSR II flow cytometer) using the annexin V‐FITC apoptosis detection kit containing propidium iodide as DNA‐binding dye (eBioscience).

MTT assay for suppression of proliferation

Isolated PBMCs from T1DM patients and healthy volunteers were incubated with different concentrations of the protein chimeric molecules in 24‐well culture plates for 3 days, as described above. The same treatment with pure anti‐CD35 (clone 3D9) antibody was also performed. Cells stimulated with a 10μg mix of both peptides (5 μg p121–140 + 5 μg p270–289) or with plate‐bound anti‐human CD3/CD28 antibodies (2·5 µg/well per each), 5 μg/ml lipopolysaccharide (LPS) (from Escherichia coli; Sigma) or cultured in medium only were used as controls.

Next, the treated PBMCs were separated into two 96‐well culture plates. Cells from the first plate were incubated for additional 4 h in the presence of MTT [3‐(4,5‐dimethylthiazol‐2‐yl)‐2,5‐ diphenyltetrazolium bromide]. After medium aspiration, dimethyl sulfoxide was added to each well and the absorbance of the dissolved formazan crystals was measured at 590 nm with background subtraction at 620 nm.

Enzyme‐linked immunospot (ELISPOT) assay

The second culture plate was used for counting of the specific anti‐GAD65 IgG antibody‐secreting plasma cells. An ethanol‐activated ELISPOT 96‐well plate (Millipore, Bedford, MA, USA) was coated with 10 μg mix of GAD65 peptides (5 μg p121–140 + 5 μg p270–289) ON at 4^oC and blocked with 1% gelatin in PBS. Next, the preincubated PBMCs from the second plate were transferred to the peptide‐coated ELISPOT plate and the cells were further cultured for additional 4 h in a humidified 5% CO₂ atmosphere at 37°C. After washing, the ELISPOT plate was incubated with an alkaline phosphatase‐conjugated anti‐human IgG antibody for 1 h and developed by nitro blue tetrazolium/5‐bromo‐4‐chloro‐3‐indolyl‐phosphate (NBT/BCIP) substrate (Sigma). The number of plasmocytes producing IgG anti‐GAD65 antibodies was detected as colored spots on the membrane counted by CTL ImmunoSpot S5 Versa Analyzer (Bonn, Germany).

Statistical analysis

All statistical analysis was performed with GraphPad Prism version 5 software (San Diego, CA, USA). The two‐way analysis of variance (anova) test was used to determine the differences between each of the two groups. Values in the figures correspond to mean ± standard deviation (SD). A value of P < 0·05 was considered statistically significant.

Results

In‐silico selection of CD4 T cell epitopes from GAD65

Four known CD4 T cell epitopes from GAD65 22 were selected based on the following filters: 20‐mer, MHC: DQ8 or DR, type of T cell response: protein‐immunized HLA transgenic mice 18. The selected 20‐mer peptides were: 121–140, 201–289, 270–289 and 556–575. They were searched for 9‐mer binders to DQ8, DRB1*04:01 and DRB1*04:05 using EpiDOCK, as described in the Methods. Peptide 121–140 contained seven DQ8 and one DRB1*04:05 binders; peptide 201–289 six DQ8 binders; peptide 270‐289 six DQ8 and two DRB1*04:05 binders; and peptide 556–575 seven DQ8 binders. The peptides 121–140 and 270–289 were selected for the construction of protein chimeric molecules. The selected potential epitopes are visualized on the GAD65 structure (Fig. 1a).

Figure 1.

The constructed GAD65 chimeras bind selectively to pathological GAD65‐specific B cells and cross‐link their BCR and CD35 on the cell surface. (a) Structure of human GAD65 (pdb code: 2okk). Epitope 1 (p121–140) is shown in blue. Epitope 2 (p270–289) is shown in red. (b) The co‐cross‐linking of BCRs and the inhibitory receptor on disease‐associated autoreactive B lymphocytes provides negative feedback regulation for B cell activation.

Construction and characterization of the protein chimeric molecules

The newly generated chimeric molecules (GAD65 chimeras 1 and 2 and control chimera) were engineered by chemical conjugation of several copies of the synthetic peptides p121–140, p270–289 or irrelevant peptide to the monoclonal anti‐human CD35 antibody (clone 3D9) (Fig. 1b). We then tested the functional activity of both elements of the designed chimeric constructs. The existing free carboxyl groups on the surface of the antibody molecule are a potential target for interaction with the reactive H₂N group of the Ahx linker bound to the C‐end of the synthetic peptides, but only a few of these carboxyl groups are available. Some of them are within the antigen‐binding sites and, as a result of peptide conjugation, the antibody affinity could be affected. To exclude this possibility, we performed competitive FACS analysis of the binding capacity of the chimeric molecules to targeted receptor on the human B and T cells isolated from T1DM patients and healthy donors compared to FITC‐conjugated anti‐CD35 antibody (clone 3D9). The GAD65 chimeras 1 and 2 retained their antigen‐binding activity and inhibited preferentially the binding of a FITC‐labeled anti‐CD35 antibody to the gated CD19⁺ cells in both patients and healthy donors. This was not so visibly manifested in the case of CD3⁺ cells, nor when the unconjugated anti‐CD35 monoclonal antibody was used (Fig. 2a,b).

Figure 2.

The GAD65 chimeras 1 and 2 retain their capability to recognize targeted receptors and to be recognized by anti‐GAD65 IgG antibodies. (a) GAD65 chimeras 1 and 2 bind to the inhibitory receptor CD35 expressed on PBMCs from T1DM patients or healthy donors. The cells were incubated with the GAD65 chimera 1, with the GAD65 chimera 2 or with control chimera, or with unconjugated 3D9 monoclonal aintibody or PBS alone for 30 min at 4°C followed by incubation with FITC‐conjugated anti‐human CD35 (clone3D9) combined with anti‐human CD19‐PE or with anti‐human CD3‐PE/Cy5 antibodies. After washing, the cells were analyzed by flow cytometry. Data are representative of at least five independent experiments. The extracted results are presented graphically. (b) The data are represented as mean ± standard error of the mean (SEM) (n = 3) (**P < 0·01, ***P < 0·001). (c) Enzyme‐linked immunosorbent assay (ELISA) for epitope availability. GAD65 peptides 1 and 2 coupled to the anti‐CD35 antibody are recognized by anti‐GAD65 antibodies. The loaded GAD65 chimera 1, GAD65 chimera 2 and control chimera were incubated in immune plates with sera from T1DM patients or healthy donors. Further, the plates were washed, incubated with alkaline phosphatase‐conjugated anti‐human IgG and developed. (d) Peptide epitope recognition by antibodies in the individual sera from healthy donors (left) and T1DM patients (right). All samples were triplicated and average values were used for analysis. Mean ± standard deviation (SD) values were calculated for each group; P‐values were calculated using the two‐way ANOVA test (*P < 0·05) in comparison to healthy donors.

Next, we studied the accessibility of the immunoglobulin‐conjugated GAD65 epitopes for interaction with the anti‐GAD65 antibodies. The GAD65 chimera 1, GAD65 chimera 2 or control chimera were loaded onto ELISA plates and further incubated with diluted sera from T1DM patients or healthy donors. GAD65‐derived peptide epitopes from both chimeric molecules were recognized by IgG anti‐GAD65 patients' sera antibodies, while the recognition from healthy donors' sera was with low intensity (Fig. 2c,d). Irrelevant peptide from control chimera was not recognized by either IgG antibodies in sera of patients or healthy donors.

GAD65 chimeras increased the percentage of apoptotic B and T lymphocytes from T1DM patients

Isolated PBMCs from clinically relevant T1DM patients and healthy volunteers were used to investigate the pro‐apoptotic effect of GAD65 chimeras. The cells were co‐cultured with the chimeric molecules for 3 days, and subsequently the surface expression of phosphatidylserine on the CD3 or CD19‐gated lymphocytes was analyzed by flow cytometry. The basic level of apoptosis of untreated B or T cells from healthy donors was much lower than registered within the patients. Increasing dose‐dependent levels of apoptosis were observed within CD19‐gated PBMCs from T1DM patients incubated with GAD65 chimeras 1 and 2, while the increased apoptotic effect within CD3‐gated cells from patients was not dose‐dependent (Fig. 3). Furthermore, the chimeras‐treated healthy donors' B cells were not affected, while an increase of apoptosis with the highest concentration of GAD65 chimera 1 was found within gated T cells isolated from the same healthy volunteers.

Figure 3.

Apoptosis of B and T lymphocytes induced by chimeric molecules. Isolated PBMCs from type 1 diabetes mellitus patients and healthy individuals were treated with increasing concentrations of GAD65 chimera 1, GAD65 chimera 2 or cultured in medium alone for 3 days. Then, the cells were stained with annexin V‐FITC/PI and analysed by flow cytometry. The plot graphs show the percentage of double‐positive cells within gated B cell (a) or T cell (b) population. The extracted results for the late apoptosis of CD19⁺‐ and CD3⁺‐treated cells are presented graphically (c). The data are represented as mean ± (SEM) (n = 3) (*P < 0·05). One typical experiment from the five performed is shown.

The control chimera did not have a significant pro‐apoptotic effect on T and B cells either from patients or healthy donors.

The GAD65 chimeric molecules suppressed the B cell differentiation to anti‐GAD65 antibody‐secreting plasma cells

The number of spots corresponding to plasmocytes secreting GAD65‐specific IgG antibodies was evaluated in an ELISPOT assay. Increasing concentrations of the constructed protein chimeras were added to PBMCs from T1DM patients and healthy individuals and the cells were co‐cultured in complete RPMI‐1640 medium without B and T cell stimulation. The treatment of PBMCs from patients with GAD65 chimeras decreases significantly the number of IgG anti‐GAD65 antibody‐secreting plasma cells (Fig. 4a). The strongest reduction of the number of plasmocyte‐secreting antibodies with this specificity was observed after incubation of PBMCs from patients with 40 and 100 ng/ml GAD65 chimeric molecules 1 and 2. The same experiments were performed using PBMCs isolated from healthy donors as controls. The background of lower number GAD65‐specific plasmocytes was not affected by the chimeric molecules treatment.

Figure 4.

ELISPOT and MTT assays. (a) The resulting spots corresponding to the number of anti‐GAD65 IgG antibody‐secreting plasma cells was determined by ELISPOT assay. Isolated PBMCs were co‐cultured with the GAD65 chimeras 1 and 2, with the control chimera or with pure anti‐CD35 antibody and each donor (patient or healthy volunteer) was tested individually. Control cells were stimulated with GAD65‐derived peptides or with lipopolysaccharide (LPS) or cultured in medium only. The number of spots in the test wells from diabetes patients (left panel) or healthy donors (right panel) were compared to untreated cells (medium only). (b) The GAD65 chimeric molecules do not inhibit the PBMC proliferation in vitro. PBMCs from T1DM patients (left panel) and healthy volunteers (right panel) were cultured in the presence of different concentrations of chimeric molecules or pure anti‐CD35 antibody. Control samples were stimulated with plate‐bound anti‐human CD3/CD28 antibodies or with GAD65‐derived peptides, or cultured in medium only. The cell proliferation was evaluated by MTT assay and the results of the test samples were compared to the untreated cells. Results are expressed as the mean value ± SD of triplicate assays. P‐values were calculated using the two‐way analysis of variance (anova) test (*P < 0·05; **P < 0·01; ***P < 0·001). Data are representative of six independent experiments.

Chimeric molecules do not influence cell proliferation in vitro

In order to investigate the effect of the treatment with GAD65 chimeras on the inhibition of cell proliferation, a colorimetric MTT assay was used. PBMCs from T1DM patients and healthy donors without stimulation were cultured in vitro with different concentrations of GAD65 chimeras 1 and 2, control chimera or unconjugated anti‐CD35 antibody for a 3‐day culture period. The control cells, isolated from both healthy volunteers and from patients, showed a strong response to the anti‐CD3/CD28 antibody stimulation while the peptide‐stimulated cells kept the same proliferation levels as unstimulated PBMCs. Under in vitro conditions the anti‐CD35 antibody alone or as part of chimeric molecules did not influence significantly the cell proliferation neither from patients nor from healthy donors' PBMCs (Fig. 4b).

Discussion

Although T1DM is a common disorder around the world, little is known about the cause of this condition. Genetic linkage with HLA on chromosome 6 and environmental triggers, such as viral infections, are among the most related to T1DM 22. It is well known that CD8 T cells have a leading role in the development of diabetes. The supportive role of CD4 T cells in the autoreactive activation of immune cells and the consequent destruction of pancreatic beta cells is suggested through the non‐obese diabetic (NOD) mouse model 7, 32. The potential role of B cells has recently attracted attention through their function not only as antibody secreting plasma cells, but as effective antigen‐presenting cells. In autoimmune diabetes, activated autoreactive B cells modify T cell regulatory function, thus breaking the T cell tolerance 17. Several studies showed that the presentation of GAD65 T cell epitopes, built into B cell epitopes, was blocked or inhibited while the T cell epitopes without antibody overlaying were presented more efficiently 21, 23. The suggestion is that the masking of T cell epitopes in the antibody‐binding region by GAD65‐specific autoantibodies stabilizes and sets apart these epitopes, thus suppressing their presentation.

By using B cell‐specific surface molecules, it may be possible to affect their development. The monoclonal antibody rituximab has been used in the treatment of autoimmune disorders such as rheumatoid arthritis and systemic lupus erythematosus (SLE). It recognizes CD20 molecules on the surface of B cells and induces their depletion by Fc receptor and complement‐mediated lysis 19, 33, 34. Although these treatments have a beneficial effect for some patients with autoimmune diseases, CD20 is absent on the surface of the B cell precursors and long‐lived plasma cells 35.

Suppression of B cell activation and proliferation by using the anti‐CD22 mAb epratuzumab, combined with blocking of the costimulatory receptors by Lympho‐Stat‐B, is also a possible solution. Unfortunately, positive data from long‐term therapy are still not reproducible 36, 37.

B cell‐activating factor (BAFF) levels are elevated in the blood of patients with rheumatoid arthritis and SLE. A proliferation‐inducing ligand (APRIL) is also suspected to be responsible for the B cell proliferation and development. The combined blocking of the BAFF and APRIL receptors by generated antibodies is currently under investigation 20, 38, 39.

All these approaches are not specific, as they deplete all the B cells and consequently leave the patients in a partly immune‐deficient condition. A possible mechanism for specific B cell down‐regulation is to cross‐link surface immunoglobulin receptors (B cell antigen receptor; BCR) with inhibitory B cell receptors using generated protein chimeric molecules.

During the autoimmune process progression, the GAD65 molecule is the most frequently attacked target by the autoreactive lymphocytes 40. The function of GAD65 in the pancreatic beta cells is still unknown. It is supposed that its product, GABA, may act as a functional regulator of pancreatic hormone release or as a paracrine signaling molecule in the communication between different cells in the islets of Langerhans 41. In different studies GABA is considered as an inhibitor or as an inducer of insulin secretion 42, 43.

The structural analysis of the enzyme has delineated several functionally important surface epitopes, suggesting that GAD65 may act as a positive T and B cell modulator, contributing to their aberrant activation and facilitating the development of autoimmune disease 44.

We have recently shown that it is possible to suppress selectively the pathological human B cells in humanized models of SLE and house dust mite allergy (HDM) by exposing them to chimeric molecules that cross‐link their inhibitory CD35 with immunoglobulin receptors. The first chimera was constructed by coupling copies of the dsDNA‐mimicking peptide to a mouse anti‐human CD35 monoclonal antibody. Its administration to humanized SCID mice reconstituted with PBMCs from SLE patients resulted in the suppression of IgG anti‐DNA antibody‐producing plasma cell, proteinuria and glomerular deposition of human IgG immune complexes 27. The second chimeric molecule contained copies of the Der p 1‐derived peptide sequence (p52–71) from Dermatophagoides pteronyssinus coupled to the same anti‐human CD35 antibody. We have shown that Der p 1‐peptide chimeric molecules cross‐linked Der p 1‐specific BCR with CD35 on the allergen‐specific B cells from HDM allergy patients, resulting in the significant reduction of anti‐Dpt IgE antibody‐producing plasma cells 28.

Here, we explore the possibility to suppress GAD65‐specific disease‐associated lymphocytes involved in autoimmune diabetes development by protein chimeric molecules consisting of GAD65‐derived peptide epitopes coupled to a monoclonal antibody specific to the human CD35. The counted CD35‐positive B cells in the PBMCs from T1DM patients was lower compared to healthy donors. B cell abnormality, overactivation and lymphopenia are typical for many autoimmune disorders such as systemic lupus 45 and T1DM 46. Several immunophenotypical peripheral B cell subset differences have been observed between T1DM patients and healthy controls. T1DM patients had fewer activation marker‐expressing B lymphocytes in some fractions and fewer class‐switched cells in mature B cell subsets than control individuals 20, 46. The authors claim that peripheral B cell maturation is disrupted in up to 50% of patients with long‐standing T1D, which could be the reason for the observed high apoptosis rate of untreated B cells in T1D patients compared to healthy donors. A similar high apoptosis level was found in B cells from the periphery in the NOD mouse model of T1DM 47.

The constructed chimeric molecules are able to cross‐link the surface immunoglobulin receptors on GAD65‐specific B cells with the inhibitory CD35 delivering a suppressive signal to the targeted cells. In‐silico generation of these potential epitopes provides a selective approach for the suppression of the existing low number of GAD65‐specific B lymphocytes without affecting the entire B cell population. A higher recognizing avidity to the combination of the targeted surface molecules (the GAD65‐specific BCRs and the CD35 receptors) is expected from chimeric antibodies resulting in preferential binding to GAD65‐specific B cells and efficient receptor co‐ligation. Indeed, both GAD65‐peptide chimeras bind preferentially to CD19⁺ B cells isolated either from patients or healthy donors, but the inhibition effect to CD35‐FITC binding (as % bound cells) was stronger when the GAD65‐specific B cells were targeted among patients' PBMCs. This was confirmed by GAD65‐epitopes recognition from serum autoantibodies from patients compared to healthy donors' sera.

The co‐culturing of PBMCs from T1DM patients or healthy donors and GAD65 chimeras 1 and 2 had a significant impact on the apoptosis levels (annexin V/PI‐double‐positive) and suggests a specific interaction of patients' B cells, while B cells from healthy donors were not affected. Neither chimera showed a high binding activity to T lymphocytes, but an increased percentage of apoptosis was observed within CD3‐positive cells isolated from both patients and healthy donors.

While the effect on patients' T cells could be explained by a restricted antigen presentation provided from suppressed GAD65‐specific B cells, the apoptosis within T cells from healthy donors may be a result from unspecific binding of CD35‐containig protein chimeric molecules during in‐vitro conditions.

The GAD65 chimeric molecules could affect autoimmune diabetes associated B and T lymphocytes by several ways: by blocking B cell differentiation to plasma cells, suppression of the plasma cell functions, inhibition of antigen presentation from GAD65‐specific B cells or restriction of B/T cell communication. Indeed, the exposure of PBMCs from T1DM patients to the GAD65 chimeric molecules 1 and 2 significantly reduced the number of IgG anti‐GAD65 antibody‐producing plasma cells but did not suppress the total cell proliferation. Under in vitro conditions the anti‐CD35 antibody possesses the ability to interact with any CD35‐expressing targets. The unbound anti‐CD35 antibody‐containing chimeras, as well as the unconjugated anti‐CD35 antibodies, are distributed to all available CD35, expressed by several groups of cells. This could explain the weak non‐significant fluctuation of cell proliferation observed after the incubation of PBMCs from either patients or healthy donors with the anti‐CD35‐containing chimeras or unconjugated anti‐CD35 antibody. The very low doses of chimera treatment have no potency for unspecific total suppression of cell proliferation, which was supported by the performed MTT assay. As a beneficial result, both GAD65 chimeric molecules, together with the control chimera, have not exhibited any unspecific suppressive effect on the total lymphocyte population. Data obtained from the ELISPOT assay showed that the targeted GAD65‐recognizing B cells are affected, while no influence to healthy donors' B cells was observed. We could expect more advantageous binding from the GAD65 chimeric molecules to their relevant target, GAD65‐specific B cells, only under in‐vivo conditions through their bivalent specificity. These chimeric molecules comprising two GAD65 epitopes cannot affect the whole spectrum of self‐reactive lymphocytes during T1DM development, and their specific suppression is restricted to B cells with these epitope specificities. Precise target mapping and self‐epitope determination could solve the future task of multi‐epitope chimeric molecule construction.

The present paper explores a novel approach for suppression of disease‐associated B lymphocytes during T1DM conditions. We demonstrate the use of a protein‐engineered chimeric molecule to down‐regulate GAD65‐associated autoreactive B cells and thus provide a possible specific therapy for human autoimmune diabetes.

Disclosures

The authors declare no conflicts of interest.

Author contributions

I. M. wrote the manuscript and researched the data. G. B. researched the data. I. D. researched the data and contributed to the discussion. N. M. researched data and reviewed/edited the manuscript. A.T. wrote the manuscript. A. T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.